For example, with the spread of server virtualization technology and network virtualization technology, networks have been complicated. Server virtualization technology is a technology in which a plurality of computers are virtually formed in a single server and an individual operating system (OS) and applications are implemented in each computer. Network virtualization technology is a technology in which various network functions are virtually formed in a server or the like using applications.
Furthermore, in addition to the above-described virtual machine (VM), for example, use of open source software (OSS), linkage between new and old systems, or the like, is desired to be employed, networks are even more complicated. When a network is complicated, a potential problem might arise while the network is in operation and influence a communication service that is being provided on the network.
In contrast, software that executes, for example, an automatic test of inputting test traffic, causing an occurrence of a pseudo failure, or the like on a network in operation has been studied and developed. Examples of such software include, for example, a product of open source software “Chaos Monkey”, for which Netflix (registered trademark) Inc. plays a central role in the development. Using “Chaos Monkey”, a virtual machine that is a test target is selected at random from a registered virtual machine group on a cloud, stop of the virtual machine or the like is forcibly executed, and the performance of restoration from a failure may be tested.
Related techniques are disclosed in, for example, Japanese Laid-open Patent Publication No. 2002-182944 and Japanese Laid-open Patent Publication No. 2005-301438.
When a test is executed on a network in operation, reduction in the quality of a communication service that is being provided on the network due to the occurrence of the above-described problem is a concern. For example, in a network that is in operation, assuming that there are three virtual firewalls between which a load is distributed by a virtual load balancer, when a test in which a pseudo failure is caused to occur in one of the virtual firewalls is performed, as long as load distribution processing is normally performed, the other two of the virtual firewalls are operated in a fallback mode and traffic processing of the virtual firewall in which the failure occurs is distributed to the other two virtual firewalls.
If an error occurs in load distribution processing, the load of traffic processing is not distributed to the other two virtual firewalls, and traffic may be discarded. That is, the entire traffic processing performance of the virtual firewalls may be reduced to about two thirds. Furthermore, in a worse case, it is conceivable that load distribution processing itself of the virtual load balancer may be stopped and traffic processing of the other two virtual firewalls may be stopped.